Silicon Valley Bank Collapse — 48 Hours of ALM Failure

In 2020–2021, the Fed's zero-rate policy and quantitative easing flooded the VC ecosystem with unprecedented capital. Startups received massive funding rounds and deposited the proceeds at SVB, causing explosive deposit growth.

SVB deposits: $61B in 2019 → $189B in 2021 (more than 3x in just two years). Capital flooded in faster than loan demand could absorb it.

Where to put all that cash? SVB management's answer: large-scale investment in long-duration MBS and US Treasuries. The logic was simple at the time — "rates will stay low for a long time."

When accounting for bond investments, banks can choose one of two classifications: HTM (Held-to-Maturity) or AFS (Available-for-Sale). This choice determined SVB's fate.

HTM carries bonds at amortized cost — even if rates rise, unrealized losses don't show up in income statements. But the moment you sell, you must recognize the full loss immediately. AFS marks bonds to fair value, with unrealized gains/losses flowing through OCI (Other Comprehensive Income).

SVB's choice: rapidly expanded the HTM portfolio to $98.2B in 2021 (78% of total investments). This was based on management's view that rates would remain low. This decision became a fatal trap in the 2022 rate-hike cycle.

SVB's bonds had an average duration of ~5.6 years. For every 1% rise in rates, the portfolio value would fall ~5.6%. As the Fed raised rates from 0.25% to 4.5% in 2022, SVB's HTM portfolio accumulated enormous unrealized losses.

Simple math: HTM $91.3B × 5.6yr duration × rate rise of +4.25% ≈ ~$21.7B potential loss. Actual disclosed unrealized HTM loss (end-2022): $15.2B — nearly equal to SVB's total equity of $16.3B.

But thanks to HTM classification, these losses were invisible in the financial statements. SVB was technically 'capital depleted,' but looked fine on paper. The problem: the moment they needed to sell those HTM bonds, everything would be exposed.

In a typical retail bank, 30–40% of deposits exceed the FDIC insurance limit ($250,000). SVB was different: 94% of deposits were uninsured. This is because customers were overwhelmingly startups and VC funds.

The VC/startup ecosystem is a tightly connected network. When Founders Fund issued a withdrawal recommendation, it spread instantly to thousands of startups via Slack and Twitter. Traditional bank runs took days — people had to physically queue. SVB's took a single tap on a banking app.

Analogy: a traditional bank run is a long physical queue (2–3 days). SVB's was a single group chat. This structural vulnerability permanently imprinted on regulatory history the critical importance of depositor composition.

Following SVB's closure on March 10, the Fed, FDIC, and Treasury issued a joint statement on Sunday evening, March 12: all SVB depositors would be fully protected. They also introduced a new tool — the BTFP (Bank Term Funding Program) — to stabilize the banking system.

BTFP's core mechanism: accept banks' HTM bonds as collateral not at market value (~70–80 cents on the dollar) but at face value (100 cents). Effect: over $600B in potential unrealized losses across US banks were instantly converted to usable collateral.

There were criticisms: implicit subsidization of depositors over shareholders, moral hazard concerns (encouraging future risk-taking), and the creation of market expectations that even non-TBTF banks could receive protection. But it was effective at stopping contagion — markets began stabilizing from March 13 onward.